[Coin-discuss] OSI branchAndBound vs CBC

[Leo Lopes]

Thanks for the clarification.

Cheers,
Leo.

On Fri, May 22, 2009 at 3:25 AM, Stefan Vigerske
<stefan at math.hu-berlin.de> wrote:
> Hi,
>
>> Could you expand on OsiCbc not being used anymore?
>
> I remember some discussions on some list that OsiCbc is not maintained
> anymore (the last change to the .cpp file was 2 years ago), and that
> things like solving again after changing the instance might not work in
> every case.
> It probably still works when one just sets up MIP via Osi, do a solve,
> and then wants a primal solution vector. Whether that gives the best
> performance is another issue.
>
>> I have software
>> which depends on OsiCbc. Is there another MIP code that should be used
>> instead via Osi?
>
> Solving MIPs via Osi is generally not the best idea (see also the recent
> discussion at http://list.coin-or.org/pipermail/osi/2009-April/000200.html).
> If it is just solving one instance, then OsiCbc still seem to work, but
> also Symphony, Glpk, and Cplex work in my experience.
>
> Stefan
>
> --
> Stefan Vigerske
> Humboldt University Berlin, Numerical Mathematics
> http://www.math.hu-berlin.de/~stefan
>



-- 
========================================================================
Leonardo B. Lopes
Assistant Professor                                        SIE -
University of Arizona
(520)621-2342
leo at sie.arizona.edu
http://www.sie.arizona.edu/faculty/leolopes